CN110059423B - Tropical cyclone objective strength determining method based on multi-factor generalized linear model - Google Patents
Tropical cyclone objective strength determining method based on multi-factor generalized linear model Download PDFInfo
- Publication number
- CN110059423B CN110059423B CN201910330964.5A CN201910330964A CN110059423B CN 110059423 B CN110059423 B CN 110059423B CN 201910330964 A CN201910330964 A CN 201910330964A CN 110059423 B CN110059423 B CN 110059423B
- Authority
- CN
- China
- Prior art keywords
- tropical cyclone
- factor
- tropical
- linear model
- generalized linear
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Cyclones (AREA)
Abstract
The invention provides a tropical cyclone objective strength determining method based on a multi-factor generalized linear model, which comprises the following steps of: step 1, extracting characteristic factors of the tropical cyclone at each moment, wherein the characteristic factors comprise MMV, RD, S-20, TBBstd, lat or Lon strength factors; and 2, establishing a relation between the characteristic factors and the maximum surface wind speed of the tropical cyclone by using a generalized linear model. The invention can further improve the accuracy of objectively determining the tropical cyclone and provides technical support for monitoring and early warning of the tropical cyclone.
Description
Technical Field
The invention relates to the technical field of tropical cyclone strength monitoring and early warning, in particular to a tropical cyclone objective strength determining method based on a multi-factor generalized linear model.
Background
Tropical cyclone is a strong cyclonic circulation which occurs on low-latitude ocean surfaces and is often accompanied by disastrous weather such as storm, rainstorm, storm and the like, and seriously threatens the life and property safety of human beings. The statistics of the world meteorological organization show that tropical cyclone has become the natural disaster causing the greatest loss to the human society in recent years, and meanwhile, china is one of the countries affected by the tropical cyclone and having the most serious disasters in the world.
In the last decade, the path forecasting capacity of tropical cyclone is remarkably improved, however, in the forecast result of tropical cyclone strength in the same time period, the tropical cyclone is wandered in the last decade, and even the forecasting precision error tends to increase year by year. In addition, the overall performance of subjective strength prediction based on the statistical prediction method still precedes the objective prediction result based on the numerical model. Therefore, the strength of the tropical cyclone remains an important issue that is currently difficult to solve.
Currently, the most widely used method for analyzing tropical cycloids based on satellite cloud charts, the Davorak technology proposed for Davorak in the last 70 th century, and various improvements that are continuously developed later. However, these methods are subjective analysis methods, require continuous and intensive analysis by the analyst, and the accuracy depends heavily on the analyst's forecasting experience and forecasting level.
Aiming at the limitations of the previous subjective analysis method,
an objective method for obtaining the shape and dynamics of TC cloud structure using infrared satellite cloud atlas, namely declination variance technique (DAV-T), is proposed. As tropical cyclonic systems develop and strengthen from unstructured clouds, the cloud structure becomes more axisymmetric, either with respect to a particular reference point. The method can obtain the degree of symmetry of each tropical cyclone system by calculating the brightness-temperature gradient on the satellite infrared cloud picture, and quantifies the degree of organization of the tropical cyclones according to the degree of symmetry. The results show that the method is objective and effective in the whole life history process of the primary period, the development period, the maturation period and the death period of the TC system. However, past single factor strength based DAV-T still has room for improvement in strength accuracy for tropical cyclones.
Disclosure of Invention
The invention introduces a plurality of auxiliary factors on the basis of DAV-T, and can greatly improve the strength fixing precision of the tropical cyclone.
The invention aims to:
the invention aims to provide a tropical cyclone objective strength determination method based on a multi-factor Generalized Linear Model (GLM), which provides reference for an objective strength determination process of the tropical cyclone.
The technical scheme of the invention is as follows:
a tropical cyclone objective strength determination method based on a multi-factor generalized linear model comprises the following steps:
step 1, extracting characteristic factors of tropical cyclone strength at each moment based on satellite high-resolution observation data and historical tropical cyclones; the characteristic factors include:
MMV: a deviation angle variance minimum value of tropical cyclone;
RD: the relative distance between the center of the tropical cyclone and the minimum deviation angle variance value of the tropical cyclone;
s-20: the area of cloud cluster brightness temperature lower than-20 ℃ within the range of 50-200 km away from the center of tropical cyclone circulation;
TBBstd: standard difference of cloud cluster brightness and temperature within 100-300 km from the center of tropical cyclone circulation;
lat: the latitude of the center of the tropical cyclone;
lon: longitude of center of tropical cyclone;
step 2, establishing a relation between the characteristic factor and the maximum surface wind speed Vmax of the tropical cyclone based on a generalized linear model, and comprising the following steps:
(201) For known m sets of training data (V) i ,X 1 ,X 2 ,X p ,...X n ),p＝1,2,...,n.，
Wherein: v i The maximum wind speed on the surface of the tropical cyclone sea at the ith moment; (X) 1 ,X 2 ,X p ,...X n ) T P =1, 2.. And n. Is V i Corresponding characteristic factors;
firstly, each characteristic factor and pairwise interaction of the characteristic factors are standardized to obtain a prediction factor item based on a generalized linear model:
wherein, the single factor action item of the predictor is as follows:
two-by-two interaction terms between predictors:
the characteristic factor based on the generalized linear model and the Vmax relationship are as follows:
wherein:
for V derived from a generalized linear model max Fitting values; b is a mixture of 0 Is a constant coefficient; (x) 1 ,x 2 ,x 3 ,...x n ) T Represents each predictor that is strong for tropical cyclones;
acting as a single factor for the predictor, b i Is the coefficient thereof;
for pairwise interaction terms between predictors, b i Is the coefficient thereof; epsilon is a residual error, and the residual error accords with normal distribution with the average value of 0;
b is obtained from formula 3 0 And b i Or b ij ；
(202) And (3) obtaining a predicted value of the tropical cyclone strength by using the characteristic factor needing to be tested in formula 3.
Preferably, in formula 3, the coefficients take the following values:
has the advantages that:
the objective strength determination method for the tropical cyclone based on the multi-factor generalized linear model can further improve the accuracy of objective strength determination for the tropical cyclone and provide technical support for monitoring and early warning of the tropical cyclone.
Drawings
FIG. 1 is a flow chart of a typhoon intensity monitoring method based on MMV fitting.
Fig. 2 is a graph comparing the strength deviation distribution based on the single-factor sigmoid function and the multi-factor GLM model for different tropical cyclone strength levels. The left graph shows the deviation distribution of the intensity-determining results of the Sigmoid model for different tropical cyclone intensity levels; the right graph is the deviation distribution of the strengthening results of the generalized linear model for different tropical cyclone strength levels. Wherein A, B, C and D respectively represent four verification sets of G-2015, G-2016, G-2017 and G-all.
Detailed Description
In order to better understand the technical content of the present invention, specific embodiments are described below with reference to the accompanying drawings.
According to the attached drawing, the objective strength determination method for the tropical cyclone based on the multi-factor generalized linear model further improves the objective strength determination precision of the tropical cyclone and provides technical support for monitoring and early warning of the tropical cyclone.
The invention is realized based on the DAV-T method and a generalized linear model.
In combination with the flow shown in fig. 1, the objective tropical cyclone strength determining method based on the multi-factor generalized linear model includes:
step 1, extracting characteristic factors of the tropical cyclone at each moment, wherein the characteristic factors comprise:
(1) MMV: deviation angle variance minimum of tropical cyclone (unit: deg) 2 )；
(2) RD: the relative distance (unit: km) between the center of the tropical cyclone and the minimum value of the deviation angle variance value thereof;
(3) S-20: the area (unit: pixel) of cloud cluster brightness temperature is lower than-20 ℃ within the range of 50-200 km away from the center of tropical cyclone circulation;
(4) TBBstd: standard difference of cloud cluster brightness temperature (unit: DEG C) within 100-300 km from the center of tropical cyclone circulation;
(5) Lat: the latitude of the center of the tropical cyclone;
(6) Lon: longitude of center of tropical cyclone;
these characteristic factors serve as the emphasis factors of the present emphasis method. The intensity-determining factor also includes the interaction between the above characteristic factors.
Step 2, establishing a characteristic factor and the maximum surface wind speed (V) of the tropical cyclone by using a generalized linear model max ) The relationship between them;
(1) For known m sets of training data (V) i ,X 1 ,X 2 ,X p ,...X n ) P =1,2, n, wherein V i Maximum speed of the tropical cyclone sea surface at different times, (X) 1 ,X 2 ,X p ,...X n ) T P =1,2, n. isCorresponding respective enhancement factors. Firstly, each fixed strength factor and interaction items thereof are standardized:
wherein, the single factor item:
two-by-two interaction terms:
the generalized linear model is represented as follows:
wherein, the first and the second end of the pipe are connected with each other,
for V derived from the model max Fitting value, (x) 1 ,x 2 ,x 3 ,...x n ) T Representing predictors of strength for tropical cyclones, b 1 ，b 2 ，...，b n The coefficients representing the terms of the respective factors,
for a single factorial contribution to each predictor,
ε is the residual error for pairwise interaction terms between predictors (by default, a normal distribution with a mean of 0 is followed).
Obtaining the coefficient b of each factor term by using the model formula 1 ，b 2 ，...，b n .
(2) The scaling factor (x) that will need to be tested 1 ,x 2 ,x p ,...x n ) Substituting p =1, 2.. And n. Into the model obtained in the step (1) to obtain the final productTo the estimated tropical cyclone strength.
Wherein the values of the model coefficients are as follows:
to verify the objective strengthening effect of the invention on tropical cyclones, the model of the invention was compared to a single factor Sigmoid strengthening model established by Pineros et al (2015). The examples of tropical cyclones generated in the northwest pacific region between 2015 and 2017 were selected as research samples. And sequentially selecting the data of each year as a test set, and taking the data of the rest years as a training set to evaluate the strength-fixing effect of the model. In addition, the data of all the years are simultaneously used as a training set and a testing set to investigate the overall strengthening effect of the model on all the data. The data set names and classifications are as follows in table 1:
TABLE 1
Considering that the DAV calculation radius has a large influence on the strength determination result of the single-factor Sigmoid model and simultaneously has an influence on partial factors in the GLM model, the root mean square error (RMS) obtained by comparing the strength determination results of two types of models at every 5 grid point intervals in the range from 25 grid points (about 250 km) to 55 grid points (about 550 km) of the calculation radius with the data set of the optimal path of the chinese meteorological office (CMA-BST) is compared, and the results are as follows:
TABLE 2
From the analysis of the table above, for all calculated radii, the root mean square error of the multi-factor GLM model is greatly reduced in all test data sets compared with that of the single-factor Sigmoid model, and the strength determination effect is obviously improved.
Shown in FIG. 2 in a different wayAnd comparing the fixed strength deviation distribution based on a single-factor sigmoid function and a multi-factor GLM model under the tropical cyclone strength level. From the strength result of the Sigmoid model, the strength deviation of the group B verification set is relatively low, and the group C verification set is relatively high. When the tropical cyclone strength is lower than the TS grade, the extreme value of the strength deviation is far more than that of the other periods, and the absolute value of the maximum strength deviation is close to 60m/s. Whereas for the generalized linear model it can be seen that the tropical cyclone is<In the change process from TD to TS intensity stage, the dispersion degree of the fixed intensity deviation is stable. Although the extreme case number of the deviation is also mostly concentrated in the period of weak intensity of the tropical cyclone, the median of the fixed deviation gradually changes from positive deviation to negative deviation with the increase of the TC intensity, and the process has a remarkable continuity characteristic. And overall, most samples have specific strength errors of-20 m.s -1 . Compared with a Sigmoid model, the generalized linear model has the advantages that the distribution of the fixed strength errors is more concentrated, the overall quantity value is smaller, and the results among different verification sets are more uniform.
Although the invention has been described with reference to preferred embodiments, it is not intended to be limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention should be determined by the appended claims.
Claims (2)
1. A tropical cyclone objective strength determination method based on a multi-factor generalized linear model is characterized by comprising the following steps:
step 1, extracting characteristic factors of tropical cyclone strength at each moment based on satellite high-resolution observation data and historical tropical cyclones; the characteristic factors include:
MMV: the declination variance minimum of tropical cyclone;
RD: the relative distance between the center of the tropical cyclone and the minimum deviation angle variance value of the tropical cyclone;
s-20: the area of cloud cluster brightness temperature lower than-20 ℃ within the range of 50-200 km away from the center of tropical cyclone circulation;
TBBstd: standard difference of cloud cluster brightness and temperature within 100-300 km from the center of tropical cyclone circulation;
lat: the latitude of the center of the tropical cyclone;
lon: longitude of center of tropical cyclone;
step 2, establishing a relation between the characteristic factor and the maximum surface wind speed Vmax of the tropical cyclone based on a generalized linear model, and comprising the following steps:
(201) For known m sets of training data (V) i ,X 1 ,X 2 ,X p ,...X n ),p＝1,2,...,n.，
Wherein: v i The maximum wind speed on the surface of the tropical cyclone sea at the ith moment; (X) 1 ,X 2 ,X p ,...X n ) T P =1,2, n, is V i Corresponding characteristic factors;
firstly, each characteristic factor and pairwise interaction of the characteristic factors are standardized to obtain a prediction factor item based on a generalized linear model:
wherein, the single factor action item of the predictor is as follows:
two-by-two interaction terms between predictors:
characteristic factor and V based on generalized linear model max The relationship is as follows:
wherein:
is composed of a broad lineV obtained by sexual model max Fitting values; b 0 Is a constant coefficient; (x) 1 ,x 2 ,x 3 ,...x n ) T Represents each predictor that is strong to tropical cyclones;
acting as a single factor for the predictor, b i Is the coefficient thereof;
as pairwise interaction terms between predictors, b i Is the coefficient thereof; epsilon is a residual error, and the residual error conforms to normal distribution with the mean value of 0;
from formula 3 to yield b 0 And b i Or b ij ；
(202) And (3) obtaining a predicted value of the tropical cyclone strength by using the characteristic factor needing to be tested in formula 3.
2. The objective tropical cyclone strength determining method based on the multi-factor generalized linear model according to claim 1, wherein in formula 3, coefficients take the following values:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910330964.5A CN110059423B (en) | 2019-04-23 | 2019-04-23 | Tropical cyclone objective strength determining method based on multi-factor generalized linear model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910330964.5A CN110059423B (en) | 2019-04-23 | 2019-04-23 | Tropical cyclone objective strength determining method based on multi-factor generalized linear model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110059423A CN110059423A (en) | 2019-07-26 |
| CN110059423B true CN110059423B (en) | 2023-01-31 |
Family
ID=67320221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910330964.5A Active CN110059423B (en) | 2019-04-23 | 2019-04-23 | Tropical cyclone objective strength determining method based on multi-factor generalized linear model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110059423B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111695473B (en) * | 2020-06-03 | 2023-12-19 | 中国人民解放军国防科技大学 | Tropical cyclone strength objective monitoring method based on long-short-term memory network model |
| CN116360013B (en) * | 2023-04-04 | 2023-10-10 | 中国气象局上海台风研究所(上海市气象科学研究所) | Typhoon objective strength determination method and system with gradient wind balance |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8160995B1 (en) * | 2008-04-18 | 2012-04-17 | Wsi, Corporation | Tropical cyclone prediction system and method |
| CN107229825A (en) * | 2017-05-23 | 2017-10-03 | 浙江大学 | A kind of tropical cyclone complete trails analogy method assessed towards calamity source |
| CN107230197A (en) * | 2017-05-27 | 2017-10-03 | 浙江师范大学 | Tropical cyclone based on satellite cloud picture and RVM is objective to determine strong method |
-
2019
- 2019-04-23 CN CN201910330964.5A patent/CN110059423B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8160995B1 (en) * | 2008-04-18 | 2012-04-17 | Wsi, Corporation | Tropical cyclone prediction system and method |
| CN107229825A (en) * | 2017-05-23 | 2017-10-03 | 浙江大学 | A kind of tropical cyclone complete trails analogy method assessed towards calamity source |
| CN107230197A (en) * | 2017-05-27 | 2017-10-03 | 浙江师范大学 | Tropical cyclone based on satellite cloud picture and RVM is objective to determine strong method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110059423A (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110058328B (en) | Multi-mode combined downscaling prediction method for northeast summer rainfall | |
| CN107728234B (en) | Thunder and lightning strength value prediction method based on atmospheric electric field data | |
| Kirchner‐Bossi et al. | A long‐term perspective of wind power output variability | |
| Hong et al. | Basis for recommending an update of wind velocity pressures in Canadian design codes | |
| CN111665575B (en) | Medium-and-long-term rainfall grading coupling forecasting method and system based on statistical power | |
| CN111680870B (en) | Comprehensive evaluation method for quality of target motion trail | |
| CN110766200A (en) | Method for predicting generating power of wind turbine generator based on K-means mean clustering | |
| CN110059423B (en) | Tropical cyclone objective strength determining method based on multi-factor generalized linear model | |
| CN108133225A (en) | A kind of icing flashover fault early warning method based on support vector machines | |
| CN109255100A (en) | A kind of Urban Rain inversion algorithm based on microwave attenuation characteristic response fingerprint recognition | |
| CN110888186A (en) | Method for forecasting hail and short-time heavy rainfall based on GBDT + LR model | |
| CN106886564B (en) | Method and device for correcting NWP (non-Newtonian Web Page) wind energy spectrum based on spatial clustering | |
| CN110781458A (en) | Method for predicting surface solar irradiance based on mixed regression model | |
| CN116449331B (en) | Dust particle number concentration estimation method based on W-band radar and meteorological satellite | |
| CN110942182A (en) | Method for establishing typhoon prediction model based on support vector regression | |
| CN113836808A (en) | PM2.5 deep learning prediction method based on heavy pollution feature constraint | |
| CN117526274A (en) | New energy power prediction method, electronic equipment and storage medium in extreme climate | |
| CN109242159A (en) | A kind of novel power transmission transmission line icing prediction technique | |
| AU2019100631A4 (en) | Self-correcting multi-model numerical rainfall ensemble forecasting method | |
| CN115186923A (en) | Photovoltaic power generation power prediction method and device and electronic equipment | |
| CN114882373A (en) | Multi-feature fusion sandstorm prediction method based on deep neural network | |
| Young | Detecting and removing inhomogeneities from long-term monthly sea level pressure time series | |
| Miller et al. | A preliminary assessment of using spatiotemporal lightning patterns for a binary classification of thunderstorm mode | |
| CN116663393A (en) | Random forest-based power distribution network continuous high-temperature fault risk level prediction method | |
| Matsuyama et al. | Spatial and temporal variabilities of rainfall in tropical South America as derived from climate prediction center merged analysis of precipitation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 211101 College of Meteorology and Oceanography, National Defense University of Science and Technology, No. 60 Shuanglong Street, Jiangning District, Nanjing, Jiangsu Province Applicant after: National University of Defense Technology Address before: 211101 College of Meteorology and Oceanography, National Defense University of Science and Technology, Chinese People's Liberation Army, No. 60 Shuanglong Street, Jiangning District, Nanjing City, Jiangsu Province Applicant before: National University of Defense Technology |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |